NEW YORK (GenomeWeb) – A team led by researchers at Harvard University and the University of Minnesota have applied a mass spectrometry-based method for analyzing DNA adducts to explore links between the Escherichia coli gentoxin colibactin to colorectal cancer.
The study, published this week in Science, made use of a technique that uses high-resolution mass spec to detect both known and unknown DNA adducts. According to Silvia Balbo, an assistant professor at the University of Minnesota, the method enables an untargeted "adductomics" approach to studying DNA adducts, somewhat analogous to proteomic or metabolomic techniques.
A DNA adduct is a modification of DNA caused by exposure to a carcinogens. They are of interest in cancer research as markers of exposure to carcinogenic materials and potentially as risk markers for the development of cancer.
In the case of the Science study, the researchers were exploring the adduct colibactin, which is produced by E. coli strains containing a gene cluster coding for a nonribosomal peptide synthetase–polyketide synthase (the pks island) that has been shown in human cells lines and animal models to cause DNA damage and speed the growth of colorectal cancers.
Other DNA adducts of interest include nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), which indicates tobacco exposure, as well as adducts created by exposure to acetaldehyde, which stem from exposures including to alcohol and air pollution.
Known adducts can be detected using immunoassays or targeted mass spec methods, but identification of previously undetected DNA adducts has been more difficult. To address this problem, Balbo and her colleagues developed an untargeted mass spec approach that they described in a 2014 Analytical Chemistry paper.
Their method uses upfront enrichment of DNA adducts combined with high-resolution mass spec to detect adducts in an unbiased manner.
As Balbo and her coauthors noted in the Analytical Chemistry paper, the method relies on the fact that under collision-induced dissociation, DNA molecules modified by adducts lose a deoxyribose. This allows the researchers to look for this loss and, when it occurs, trigger another round of fragmentation to collect mass spec data on the adduct present.
Prior to mass spec analysis, DNA of interest is hydrolyzed to generate free deoxyribonucleosides, which are then run on a liquid chromatography platform to separate DNA molecules bound to adducts from unbound molecules. The adduct-bound molecules are then analyzed via mass spec to collect data on what adducts are present.
The technique takes advantage of technologies like nanospray ionization and data-dependent acquisition mass spec that have been largely the domain of proteomics, said Peter Villalta, mass spectrometry laboratory manager at the University of Minnesota and senior author on the Analytical Chemistry study.
"We're using instrumentation that has been improved and refined for proteomics and using the high-resolution capabilities of the some of these instruments," he said.
Prior to the University of Minnesota team's development of the approach, groups using mass spec to analyze DNA adducts typically used triple quadrupole instruments to look at known adducts in a targeted manner.
"Usually people have been taking a [carcinogen], reacting it with DNA or nucleosides, identifying what the major modification would be, and then looking in animal models or humans in a very targeted way to see whether that modification could be found," Balbo said. This approach, she noted, usually examines "one or two or a maximum of four or five adducts at a time, having characterized them structurally beforehand."
"Our approach is more agnostic, where we are potentially [identifying] adducts that have already been characterized, but also ones that have not been characterized," she said.
Villalta noted that, unlike proteomics, where experimental mass spectra can be compared to peptide spectra predicted by genome sequences to make protein identifications, the unbiased adductomics approach often produces spectra that are difficult to match to a known molecule.
While researchers can get molecular formulas and some structural information from their mass spec analyses, many of the adducts they analyze are "completely unknown," he said.
"The list of known adducts that we can search against is very primitive in that we don't have software or libraries or databases that we can search against to identify [an adduct]," Balbo said.
She and her team are working with collaborators to begin building such libraries, she said, "but we're not there yet."
Since publishing the 2014 paper, the researchers have refined the method by moving it to a higher performance mass spec (from a Thermo Fisher Scientific Orbitrap Velos to an Orbitrap Fusion) and by refining their sample prep and clean-up steps to remove impurities that interfered with the adduct spectra they aimed to interpret.
Balbo said the Science paper served as proof that the workflow they presented five years ago could be useful experimentally.
"For us it's a big deal," she said. "We suggested we could do this in our first paper and our work since then, but this is really proof that we can do it."
Emily Balskus, professor of chemistry and chemical biology at Harvard and senior author on the Science paper said she and her colleagues turned to the University of Minnesota team's adductomics approach due to the lack of knowledge around the structure of colibactin.
"We had some hypotheses about what structural features [colibactin adducts] might contain based on work we had done previously studying the biosynthetic enzymes that make the toxin, but no one has managed to isolate the intact toxin, and it's clear that there are a lot of gaps in our understanding of its structure," she said. "So we didn't want to make any assumptions about what the structure might look like and wanted to really look for adducts in a completely unbiased way."
Balskus said that while work by her lab and others had suggested colibactin might be capable of directly reacting with DNA, "people had not shown that in a biologically relevant setting."
And, in fact, their initial efforts using standard metabolomic approaches to identify differences in DNA samples from HeLa cells exposed to colibactin-producing and non-producing E. coli strains proved unsuccessful.
Using the unbiased adductomics approach, the researchers were able to show that human cell lines treated with colibactin-producing E. coli contained colibactin-DNA adducts and that metabolites derived from this adduct were present in mice infected with colibactin-producing E. coli.
Balskus said the causal link between these E. coli strains and colorectal cancer is still being developed, though multiple studies have demonstrated a correlation between the two.
"Moving forward, a big question is if we can find these adducts in patient tumor samples," she said.
Balbo said that in addition to continuing development of the adductomics methodology, she and her colleagues are using the approach to address DNA adduct questions of their own.
In addition to more comprehensively characterizing adducts produced by behaviors like smoking and drinking as well as exposure to different chemicals known to produce DNA damage, the researchers are working to characterize adducts produced by exposure to chemotherapeutic drugs.
"A lot of [chemotherapies] have interactions with DNA, and we're trying to better characterize these interactions," Balbo said, noting that this could help them better understand why patients may be more or less susceptible to side effects from treatment or more or less likely to benefit from a particular therapy.
They are also working with pharmaceutical firms using the method to complement existing genotoxic screening methods, she said.